chia7712 commented on PR #16491: URL: https://github.com/apache/kafka/pull/16491#issuecomment-2204040094
> Thus, I don't think it will cause the error as described in the JIRA, right? I still don't understand in what circumstance this issue will happen. It might need more explanation. Thanks. thanks for this response, and it inspires me to dig-in the root cause. The issue happens when the zk broker is using generated broker id and it is in migrating. 1. if we define the broker id in config file, the broker id is viewed as invalid since it is in the reserved range (when `broker.id.generation.enable=true`) 2. if we don't define the broker id in config file, the broker id will be evaluated according to `meta.properties` [0] . However, we forgot to update `node.id` and so kraft client will use "node.id=-1" [1] to build and it results in error .. [0] https://github.com/apache/kafka/blob/trunk/core/src/main/scala/kafka/server/KafkaServer.scala#L261 [1] https://github.com/apache/kafka/blob/e55c28c60b0f021b12c93fa04e360ce7a2e0a5ac/core/src/main/scala/kafka/raft/RaftManager.scala#L230 In short, we don't consider the use case of "generated broker id (the broker id is omitted)" + "zk migration". It seems we need two fixes: 1. update docs to explain the possible error when you define the broker id for the zk broker which is using `broker.id.generation.enable` 2. in order to minimize the changes, we can update both `broker.id` and `node.id` together. The side effect is `KafkaConfig#nodeId` will be a mutable variable. @showuon WDYT? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org